Most approaches to the automation of assembly tasks have suffered from the fact that their controllers do not access information about the relative positions of the parts being assembled. Without this information, real time relative position correction is prohibited.
It has been recently recognized that, in many assembly tasks, the contact force that results from part misalignment contains the required relative positional information to identify the motion that alleviates the misalignment. Many of the previous approaches to automated assembly have ignored this form of relative positioning information. Contact forces have been regulated to prevent damage, but not exploited to improve relative positioning. To realize automated assembly, the relationship between the forces applied to the assembly and the resulting motion (this relationship is the "compliance") is designed so that the forces associated with all types of positional misalignment lead to motions that cause a reduction in part misalignment, ultimately resulting in successful assembly. The appropriate compliance is selected by evaluating the geometries of the mating parts and the magnitude of contact friction.
The design of compliance for improved force guidance was first addressed by D. E. Whitney, "Quasi-Static Assembly of Compliantly Supported Rigid Parts", ASME Journal of Dynamic Systems, Measurements, and Control, 104(1), 1982. Whitney showed that, for a restrictive class of assembly problems, force guided assembly can be achieved with proper placement of a compliance center--a very small and simple class of compliance. A group at the Charles Stark Draper Laboratory developed a device that properly locates a compliance center for chamfered peg-in-hole insertions, the remote center compliance (RCC) as shown in U.S. Pat. No. 4,098,001 to Watson. This approach to compliance design has one major limitation: restricted application. Not all assembly tasks are as simple as peg-in-hole insertion.
Recently, Schimmels et al., "Admittance Matrix Design for Force Guided Assembly", IEEE Transactions on Robotics and Automation, 8(2), 1992, provided: 1) a means of identifying those tasks that can be reliably addressed using force guidance, and 2) a systematic approach to the design of a manipulator's compliance matrix that allows reliable force guided assembly (in those tasks), if friction is zero and part misalignment is small. Concepts were validated in a testbed application of inserting parts of known simple geometries into fixtures. Later, Schimmels et al., "Force-Assembly with Friction", IEEE Transactions on Robotics and Automation, 10(4), 1994 determined that reliable force guidance is achieved despite friction, if relatively complicated compliance behavior is used. This more complicated compliance behavior must provide directional coupling (J. M. Schimmels, "Multidirectional Compliance and Constraint for Improved Robot Positioning and Bracing", IEEE Transactions on Robotics and Automation), for which case, contact forces lead to motions in directions different from that of the applied force. The mechanism presented here is a means of realizing any specified compliance behavior--the compliance matrix best for a given assembly task--so that force guided assembly is achieved passively.
One area that would greatly benefit from passive force guidance is robotic assembly. Conventional position controlled manipulators are designed to be stiff, so that external forces will produce only minimal deflection of the manipulator. Yet, to achieve fast, reliable, force guided assembly, the manipulator must not only comply with contact forces, but comply is a prescribed way (by having the prescribed compliance matrix). New controller designs that allow a manipulator to achieve the appropriate compliance for a task (to behave more dexterously) are being investigated by many researchers.
Griffis and Duffy in "Kinestatic Control: A Novel Theory for Simultaneously Regulating Force and Displacement", ASME Journal of Mechanical Design, 113(4), 1991 have described the use of a 6 DOF passive parallel compliant mechanism for use in simultaneously regulating force and displacement in robotic assembly. Their device consisted of 6 elastic components that transmit only translational force along the axis of the component. Huang and Schimmels in "The Bounds and Realization of Spatial Stiffnesses Achieved with Simple Springs Connected in Parallel", IEEE Transaction on Robotics and Automation, (accepted for publication), 1997, however, showed that a mechanism of this type is only capable of achieving a very small set of general spatial compliant behavior. To eliminate this restriction, the inventors have discovered by means of the present invention, that, to achieve an arbitrary spatial compliant behavior, the elastic components connected in parallel must have helical compliant behavior. To realize this behavior, at least one of the elastic components connected in parallel must transmit force along and torque about its axis. This form of coupled translational and rotational elastic behavior is necessary in spatial assembly to achieve robust force guidance--force guidance with the capability to tolerate finite positioning errors and finite values of friction.
The present invention has evolved from initial research conducted by the inventors and set forth in Appendix I entitled "Achieving an Arbitrary Spatial Stiffness With Springs Connected in Parallel."
Using the mechanism described here, the best compliance for an assembly task could be built into a passive mechanism that is mounted on the end-effector of a conventional robot. The attractive features of the passive end-effector mounted compliant device are its simplicity, its reliability, and its speed.